Insights From Genomic Approaches to Oncology Discovery

Photo by © ASCO/Todd Buchanan 2012

Mark A. Rubin, MD

In a keynote address at the ASCO Genitourinary Cancers Symposium, Mark A. Rubin, MD, discussed work in collaboration with colleagues to develop a comprehensive molecular classification of prostate cancer (a taxonomy) using next-generation sequencing paired with novel computational tools. Rubin also brought attendees up to date on work involved in personalized medicine, or as he calls it, “precision medicine” for prostate cancer, which is still an elusive goal. Rubin is vice-chair for Experimental Research at the Weill Cornell Medical College in New York City.

“Tests and technologies are rapidly changing, leading us to think we can attain the goal of precision medicine. One of the key components of ‘precision medicine’ is a new taxonomy of human disease, following a mandate,” Rubin explained.

“The concept of looking at an individual cell is not entirely new. But newer technologies allow us to go deeper into the genome, and we anticipate unexpected findings,” he said.

Rubin hypothesized that prostate cancer is a collection of homogenous subtypes identifiable by molecular criteria; in other words, that it has underlying molecular signatures, as has been well defined in leukemias and breast cancer, but not yet in prostate cancer.

Support of this concept is the observation that some prostate cancers metastasize exclusively to the bone, while others metastasize to different sites. This may be explained by underlying molecular alterations.

“We hope that finding the underlying molecular differences will distinguish each of the different subtypes of prostate cancer,” he continued.

For at least a decade, this search has been elusive. However, in 2005, Tomlins et al reported in Science that about 50% of prostate cancers are preceded by an earlier lesion expressing ERG. Since then, using an inductive approach, a number of rearrangements involving S genes have been identified: 56% are fusion cancers, and 44% are nonfusion cancers. There is great interest in this area,” Rubin stated. “We have a starting point: ERG-arranged genes.”

Other research teams have found BRAF and KRAS rearrangements in prostate cancer, suggesting targetable subclasses due to these rearrangements. An example is the NF-kappaB pathway, which is rearranged in approximately 20% of prostate cancers.

Scientists at the Broad Institute, Dana-Farber Cancer Institute, and Weill Cornell Medical College are involved in genome-sequencing studies.

“At Weill Cornell, we are looking for driving mutations of tumors. Overall, there is a level of skepticism about this approach. It is difficult to distinguish between drivers and passengers, and we are trying to correct for this. The process entails pipeline pathology review, tissue samples, DNA extraction, and DNA analysis to identify somatic alterations,” he explained.

This research has uncovered about 180 genomes involved in a large number of prostate tumors. DNA translocations have been identified, and a new insight is that the DNA in prostate tumors has undergone a significant amount of rearrangement. “This was not previously appreciated,” Rubin noted.

ERG-rearranged tumors were found to have particular characteristics. A preliminary observation is that specific areas of the genome come together when you have ERG overexpression. “This suggests that certain rearrangements are caused because ERG brings them together,” he said. ERG-positive tumors have different patterns than ERG-negative tumors, and this has implications for DNA stability and for targeting.

“

We are navigating a paradigm shift where the goal of molecular testing is to have actionable results. We need to consider feasibility, benefits, and costs.”

—Mark A. Rubin, MD

One concept is that a transcription factor oncogene can have specific and predictable results. Unpublished data on 120 primary prostate tumors identified 5769 somatic mutations in tumor versus control, with 10-105 mutations per tumor.

“We are figuring out what to do with these data,” Rubin said.

There are recurrent mutations; for example, 4% of cases have FOXA1 in areas that bind DNA. FOXA1 is an important regulator in hormone transcription.

The most frequent mutation category is the SPOP mutation, found in 13% of prostate tumors. Other studies have confirmed that SPOP mutations occur in 6% to 13% of prostate cancers.

SPOP is different from ERG and S transcription factors. The role of SPOP is to tag specific proteins for degradation, Rubin explained. In some instances, SPOP rearrangements and ERG rearrangements are mutually exclusive.

“We are starting to see a theme that DNA repair may be related to some the regions we are finding. Several institutions are doing very exciting studies, some of them in the zebrafish model to look at genetic alterations,” he said.

“In summary for the exome work, we see SPOP as the most common mutated gene in prostate cancer, and we think it will be important,” he stated.

Turning to a precision medicine, Rubin said, “We are navigating a paradigm shift where the goal of molecular testing is to have actionable results. We need to consider feasibility, benefits, and costs. We want to be there, but we don’t know exactly where ‘there’ is. At best, we’ve found putative targets and few actionable genes. In the future, we may know every mutation, but we don’t know what we will do with these interesting but unproven alterations and consider how these mutations play a role in driving pathways in prostate cancer.”

“Precision medicine will require a team approach to make sense ethically, scientifically, and economically,” Rubin concluded.